Computer engineering grad students take 2nd place in national IBM intercollegiate challenge

Associate Professor Magdalini Eirinaki with winning Team DataSc, l-r, Professor Eirinaki, Ryan Alan Vo, Ananya Kaulgi, Soniya Hadkar.

Associate Professor Magdalini Eirinaki with winning Team DataSc, l-r, Professor Eirinaki, Ryan Alan Vo, Ananya Kaulgi, Soniya Hadkar.

It was a defining moment for three computer engineering grad students – Soniya Hadkar, Ananya Kaulgi and Ryan Alan Vo – when they learned their DataSc (Data Scientists) team had won second place in the national IBM intercollegiate competition. The Great Mind Challenge (TGMC): Watson Edition. Thirty-seven university teams, including eight from SJSU, competed in the prestigious challenge.

Watson, an artificial intelligent computer system created by IBM, first gained fame in 2011, when it outperformed human opponents on the popular television game show Jeopardy! For this challenge, students had to apply machine learning algorithms on a real dataset in order to predict the correct answers to quiz questions, mimicking one of the key processes that Watson runs during the game.

Getting started

SJSU became a contender when computer engineering Associate Professor Magdalini Eirinaki learned about the challenge and explained it to students in her Web and data mining classes last fall. She offered students the opportunity to compete in TGMC as a way to fulfill the required project assignment in the course, rather than undertaking a more traditional project.

Throughout the five-week competition process, Eirinaki was challenged right along with her students. “Initially, the online information from IBM came at a very slow pace,” she recalls, “so I wasn’t sure what to expect or how it would fit into the course syllabus. It was also a first – having a competition as part of the course. However, I was excited about the opportunity and was confident it would be a great learning experience for my students,” Eirinaki says.

As IBM shared more details, Eirinaki introduced the basic concepts and workflow of the Watson platform into her lectures, so students would have the fundamentals by the time the competition started. Since not all her students participated in the challenge, she limited the time she spent presenting details in her lectures, and set up a page in Canvas to share all the documentation. Once the competition started, Eirinaki worked with students on a team-by-team basis, providing directions and feedback on their approaches.

“I wasn’t involved in the implementation efforts, so definitely the DataSc team members deserve all the kudos for the commitment and hard work it took to place second in a nation-wide competition, and with little previous background,” affirms Eirinaki. “I’m very proud of what they’ve accomplished.”

Eirinaki is already pondering how she can leverage the success of this competition for future classes – perhaps participating in Kaggle competitions.

Team DataSc

“Initially, our team had to work extra hard getting up to speed because data from IBM wasn’t self-explanatory,” says Kaulgi. “However, Professor Eirinaki was extremely encouraging and her lectures helped filled in the gaps, solving much of the confusion.

“The aim of the project was for us to learn and innovate,” continues Kaulgi. “We had to prepare and understand many concepts associated with the project before we could start the process of executing it. It required thinking out of the box, applying techniques, and experimenting with existing methods.”

“We were very excited, but also scared because we had to produce results within five weeks,” says Hadkar. “We were challenged to try out different combinations of various mechanisms, and adhere to timelines and also maintain good coordination as a team. It was quite tricky in the beginning, but we got the hang of it.”

“Although we were unsure whether to enter the competition, we took a leap of faith even though we didn’t have a background in machine learning,” adds Vo. “We did a lot of research, including studying IBM papers on Watson, which were a bit difficult to understand. Throughout the process, Professor Eirinaki was quite supportive and believed we had a chance, even though we weren't so sure at times,” he says.

“We’re very proud of these outstanding computer engineering students,” says Dean Andrew Hsu. “This accomplishment shows the commitment and capability of our students to achieve and gain recognition on the national stage. It’s also another example of a dedicated faculty member who is student-centered and willing to go the extra mile for her students, so they can take advantage of a truly unique hands-on learning experience. All of the eight teams involved in this competition are winners.”